AMD Athlon 5350 And AM1 Platform Review: Kabini In A Socket
Last response: in Reviews comments
AMD is repackaging the mobile-focused Kabini APU for use on the desktop, and tucking them under the familiar Athlon and Sempron brands. We take a closer look at the company's socketed AM1 platform and compare its performance to Intel's Bay Trail-D.
AMD Athlon 5350 And AM1 Platform Review: Kabini In A Socket : Read more
AMD Athlon 5350 And AM1 Platform Review: Kabini In A Socket : Read more
More about : amd athlon 5350 am1 platform review kabini socket
srap
April 9, 2014 5:31:19 AM
kartu
April 9, 2014 5:37:05 AM
Related resources
AMD Radeon
April 9, 2014 5:37:28 AM
Niko_boy
April 9, 2014 5:42:45 AM
Maxx_Power
April 9, 2014 5:59:40 AM
To Toms:
What was the ambient temperature during the testing ? Surely Toms is not testing in an refrigerator ? That idle GPU and CPU temperature at 13 degrees Celsius seems about 10 degrees too low, given that they have to be necessarily above ambient conditions to make physical sense. Perhaps the sensors are not read correctly (wrong offset) ?
For the Dota2 graphs, there seems to be a mistake for the color legend. The difference in color isn't resolution, since the resolution is fixed (right hand top corner of graph at 1080), so my guess is that red/black represents min/avg FPS similar to the 2nd graph on Grid2.
What was the ambient temperature during the testing ? Surely Toms is not testing in an refrigerator ? That idle GPU and CPU temperature at 13 degrees Celsius seems about 10 degrees too low, given that they have to be necessarily above ambient conditions to make physical sense. Perhaps the sensors are not read correctly (wrong offset) ?
For the Dota2 graphs, there seems to be a mistake for the color legend. The difference in color isn't resolution, since the resolution is fixed (right hand top corner of graph at 1080), so my guess is that red/black represents min/avg FPS similar to the 2nd graph on Grid2.
Score
8
bustapr
April 9, 2014 6:13:12 AM
nezzymighty
April 9, 2014 6:14:33 AM
Thanks for the article. I agree with your conclusion, as I would probably seek an alternative build for a a "PC-Like Device" as you put it. Seeing AM1 product costs relatively the same to an A4-4000 FM2+ confuses me, as I had originally expected (hoped) AM1 high-end combinations with a motherboard to cost around the $70 mark... If I were to make an AMD HTPC, or a typical AMD business computer, based on prices released by the e-tailers today, it would make more sense for me to build an FM2+ with A4-4000. Maybe prices will reduce in time to make this a more attractive alternative relative to an FM2+ for an HTPC or business computer.
Score
4
ykki
April 9, 2014 6:28:16 AM
"Based on our results, I'd guess that AMD's AM1 platform should be able to handle less-demanding MMOs like World of Warcraft, lightweight shooters like Left 4 Dead, and a wide range of even more casual games (such as Angry Birds)."
I think that it is safe to say that it can max out angry birds at 4k
I think that it is safe to say that it can max out angry birds at 4k
Score
0
ykki
April 9, 2014 6:35:01 AM
Maxx_Power said:
To Toms:What was the ambient temperature during the testing ? Surely Toms is not testing in an refrigerator ? That idle GPU and CPU temperature at 13 degrees Celsius seems about 10 degrees too low, given that they have to be necessarily above ambient conditions to make physical sense. Perhaps the sensors are not read correctly (wrong offset) ?
On-die sensors are notoriously inaccurate at low temperatures. In this case that's obvious, but we can only report what the sensors tell us.
Maxx_Power said:
For the Dota2 graphs, there seems to be a mistake for the color legend. The difference in color isn't resolution, since the resolution is fixed (right hand top corner of graph at 1080), so my guess is that red/black represents min/avg FPS similar to the 2nd graph on Grid2.Good catch! Fixed.
Score
3
m32
April 9, 2014 6:50:51 AM
Wisecracker
April 9, 2014 6:52:52 AM
CaptainTom
April 9, 2014 7:08:43 AM
logainofhades
April 9, 2014 7:12:27 AM
Wisecracker
April 9, 2014 7:20:25 AM
zooted
April 9, 2014 7:57:08 AM
de5_Roy
April 9, 2014 8:27:01 AM
iceman26
April 9, 2014 8:43:43 AM
EdgeT
April 9, 2014 8:44:22 AM
They look like great little platforms. But a whole lot of people go the AMD route for homeservers (myself included). And just like me, a lot of people think that 4 SATA ports are just not enough. If it had like 8, like a lot of AMD motherboards I've looked at, I'd buy it in a heartbeat, and I bet a lot of people would.
It's just convenience and cost efficiency, really, on the one hand, you've got 1 motherboard, CPU, tower, PSU, OS and IP adress for remote control and 8 HDDs and on the other, you've still got 8 HDDS, but 2 of everything else. Power consumption wouldn't make THAT much of a difference, since storage servers mostly idle, but the noise and size do take their toll.
It's just convenience and cost efficiency, really, on the one hand, you've got 1 motherboard, CPU, tower, PSU, OS and IP adress for remote control and 8 HDDs and on the other, you've still got 8 HDDS, but 2 of everything else. Power consumption wouldn't make THAT much of a difference, since storage servers mostly idle, but the noise and size do take their toll.
Score
0
greghome
April 9, 2014 8:44:58 AM
iceman26
April 9, 2014 8:49:23 AM
nowadays when i'm thinking about building an HTPC i'm looking forward to use it with Madvr but seeing how madvr punishes my hd7850, i guess the two doesn't want to be together (HTPC+madvr). but i think this am1 system is for super light weight use. (e.g. downloading stuff, server) but when it comes to this kind of thing the first thing i'm concerned of was power usage .
Score
-1
EricJohn2004
April 9, 2014 8:52:45 AM
I was curious to read this because I own a PS4. And my PS4 has 8 Jaguar cores the EXACT same as these Kabini cores. So basically, if you multiply the benchmark scores by 2x, you'll get the score my PS4 would have, round about. Except for single core benchmarks, but even in that case it looks like AMD's Jaguar does pretty good with single threaded stuff. But it's not like the PS4 will have anything single threaded most likely everything will be using a ton of threads.
I also read on another website that the 5350 scores around a 1.6 in Cinebench 11.5. So it's like .4 per core, which is about half as much as an 8350 core or 4300 core. Meaning my PS4 would score a 3.2 in Cinebench 11.5. Being that that's what an FX4300 scores, i'd say a PS4's CPU is really no slouch, especially while multithreading. Sure, if you compare it to an i7, an i7 will double it's scores and more, but being that it matches an FX4300 and A10 7850k's, I'd say that's impressive.
Of course it's only half as fast as FX4300 in single threaded tasks, but that's good enough.
Any of you impressed?
I also read on another website that the 5350 scores around a 1.6 in Cinebench 11.5. So it's like .4 per core, which is about half as much as an 8350 core or 4300 core. Meaning my PS4 would score a 3.2 in Cinebench 11.5. Being that that's what an FX4300 scores, i'd say a PS4's CPU is really no slouch, especially while multithreading. Sure, if you compare it to an i7, an i7 will double it's scores and more, but being that it matches an FX4300 and A10 7850k's, I'd say that's impressive.
Of course it's only half as fast as FX4300 in single threaded tasks, but that's good enough.
Any of you impressed?
Score
0
EricJohn2004
April 9, 2014 9:07:39 AM
Yes, I was about right, the FX4300 scores a 3.13 and an i3 3225 scores a 3.25. So PS4's CPU is a little faster than FX4300 and a tad slower than i3 3225.
Personally, I thought this CPU would be a LOT slower. Like maybe it would score a 1.5 in Cinebench 11.5. Since everyone said it was a low power tablet CPU. Turns out that that's far from the truth.
And being that you get a GPU that has the same amount of cores and memory bandwidth as a GTX760. I'd say the PS4 is a damn good deal at only 399$.
Of course I own a top of the line PC too with an i5 3570k@4.7Ghz and a GTX780@1200Mhz with a 500GB SSD and a 27 inch 1920x1080 IPS screen. And that's awesome and all. I just prefer to play single player games like Tomb Raider, and Infamous Second Son on my couch where it's not as competitive, and I play multiplayer games like BF4 and World of Tanks on my PC, where it's a lot more competitive.
I'm just glad that the PS4 has powerful enough hardware to at least end up matching mid/high end GTX760 as devs get used to it's power. Because on PC a GTX760 almost max's out every game at 1080p 60FPS. So there's no reason a PS4 shouldn't be able to do it.
Personally, I thought this CPU would be a LOT slower. Like maybe it would score a 1.5 in Cinebench 11.5. Since everyone said it was a low power tablet CPU. Turns out that that's far from the truth.
And being that you get a GPU that has the same amount of cores and memory bandwidth as a GTX760. I'd say the PS4 is a damn good deal at only 399$.
Of course I own a top of the line PC too with an i5 3570k@4.7Ghz and a GTX780@1200Mhz with a 500GB SSD and a 27 inch 1920x1080 IPS screen. And that's awesome and all. I just prefer to play single player games like Tomb Raider, and Infamous Second Son on my couch where it's not as competitive, and I play multiplayer games like BF4 and World of Tanks on my PC, where it's a lot more competitive.
I'm just glad that the PS4 has powerful enough hardware to at least end up matching mid/high end GTX760 as devs get used to it's power. Because on PC a GTX760 almost max's out every game at 1080p 60FPS. So there's no reason a PS4 shouldn't be able to do it.
Score
0
tourist
April 9, 2014 9:20:08 AM
There are some people who bought E-350 or E-450 machines and were disappointed by their performance. This is the chip it looks like they needed. Still weak, but perhaps enough. I too would like to see some tests with a discrete GPU. Maybe no one would buy them together, but how many of us who got a chance to test one of these wouldn't pull an old HD4850 or HD7750 out of the parts box?
Did you test the 19V input? That is HUGE, making this perfect for a kid's entertainment machine on the road (someone just needs to sell the monitor for it...hint hint...).
Did you test the 19V input? That is HUGE, making this perfect for a kid's entertainment machine on the road (someone just needs to sell the monitor for it...hint hint...).
Score
0
Kai Dowin
April 9, 2014 10:16:47 AM
xiinc37
April 9, 2014 10:27:22 AM
rwinches
April 9, 2014 10:30:40 AM
rwinches
April 9, 2014 10:34:49 AM
FuzzyIce
April 9, 2014 11:25:55 AM
ieman26, if by a download only machine you mean something to hook to the router for download stuff from the Internet, this would be a very good choice. I just saw an ad offering a mini ITX motherboard with 4 SATA 6Gbps and the Sempron plus a mini atx case for $99. You can easily put together a small footprint PC with up to 4 sata hds and download anything you want. You can make it a Linux machine, for the sake of being protected from virus and malware, and you could run it headless, and use another PC in your network to remote login there and schedule all the downloads for you. Hope this helps.
Score
0
EdgeT
April 9, 2014 11:39:30 AM
They look like great little platforms. But a whole lot of people go the AMD route for homeservers (myself included). And just like me, a lot of people think that 4 SATA ports are just not enough. If it had like 8, like a lot of AMD motherboards I've looked at, I'd buy it in a heartbeat, and I bet a lot of people would.
It's just convenience and cost efficiency, really, on the one hand, you've got 1 motherboard, CPU, tower, PSU, OS and IP adress for remote control and 8 HDDs and on the other, you've still got 8 HDDS, but 2 of everything else. Power consumption wouldn't make THAT much of a difference, since storage servers mostly idle, but the noise and size do take their toll.
It's just convenience and cost efficiency, really, on the one hand, you've got 1 motherboard, CPU, tower, PSU, OS and IP adress for remote control and 8 HDDs and on the other, you've still got 8 HDDS, but 2 of everything else. Power consumption wouldn't make THAT much of a difference, since storage servers mostly idle, but the noise and size do take their toll.
Score
1
ojas
April 9, 2014 11:54:17 AM
lp231
April 9, 2014 12:31:18 PM
Quad core Celerons, interesting. Basically Intel can release whatever CPU they want if they felt like it, like a unlock Haswell Pentium.
http://www.tomshardware.com/news/intel-haswell-e-devils...
If they can do that, they should also release some 6 cores for the current Haswell 1150.
http://www.tomshardware.com/news/intel-haswell-e-devils...
If they can do that, they should also release some 6 cores for the current Haswell 1150.
Score
0
Jaroslav Jandek
April 9, 2014 1:02:25 PM
Gotta love that 850W PSU for testing of 30W parts - the true meaning of overkill
. I would like to see them tested with those mini PSUs too (even though the difference should be negligible).
Also, there is an error in Productivity Benchmarks - Photoshop encoding graph: "higher is better" should be "lower is better".
I am actually planning to buy a mini-PC, so I looked the prices up. Most "mainstream" shops have them more expensive, though... like newegg has A5350 for $65:
GA-AM1M-S2H ($42) + Athlon 5350 AM1 R3 ($60) = $102.
I can get Asrock Q1900B-ITX J1900 MITX for $84.
I am probably going to wait a bit and buy a 14nm NUC next year, since I don't really need that mini PC (the room I am placing it in isn't finished anyway).
. I would like to see them tested with those mini PSUs too (even though the difference should be negligible).Also, there is an error in Productivity Benchmarks - Photoshop encoding graph: "higher is better" should be "lower is better".
kartu said:
Why is there no pricing on Intel CPUs on page 1?GA-AM1M-S2H ($42) + Athlon 5350 AM1 R3 ($60) = $102.
I can get Asrock Q1900B-ITX J1900 MITX for $84.
I am probably going to wait a bit and buy a 14nm NUC next year, since I don't really need that mini PC (the room I am placing it in isn't finished anyway).
Score
2
Drejeck
April 9, 2014 1:10:01 PM
killerclick
April 9, 2014 1:12:00 PM
Zepid
April 9, 2014 1:16:12 PM
ingtar33
April 9, 2014 1:28:04 PM
crisan_tiberiu
April 9, 2014 2:03:08 PM
Nintendo Maniac 64
April 9, 2014 2:15:37 PM
Daniel Bayes
April 9, 2014 2:26:42 PM
The power testing is completely wrong here for a few reasons:
1) No one in their right mind will run an AM1 or Celeron CPU on an 850W PSU.
2) The efficiency of PSUs drops significantly at low (under 10%) utilization.
3) PSUs have a small overhead when running, which is significant at extremely low wattages.
4) 80+ certification says absolutely nothing about running at around 5-7% of the maximum PSU output - even with platinum certification, at those percentages it is usually far less than 70% efficiency.
A far better test would be a 70-100W PSU, which would give a result that is both far more accurate and in this case, crucially important when you cite it as a reason to give AMD such an edge in recommendation.
1) No one in their right mind will run an AM1 or Celeron CPU on an 850W PSU.
2) The efficiency of PSUs drops significantly at low (under 10%) utilization.
3) PSUs have a small overhead when running, which is significant at extremely low wattages.
4) 80+ certification says absolutely nothing about running at around 5-7% of the maximum PSU output - even with platinum certification, at those percentages it is usually far less than 70% efficiency.
A far better test would be a 70-100W PSU, which would give a result that is both far more accurate and in this case, crucially important when you cite it as a reason to give AMD such an edge in recommendation.
Score
0
compdoc
April 9, 2014 3:00:37 PM
I've been building routers with kabini for a couple of months now thanks to ECS KBN-I boards. And now I have more choices. These systems are great for running pfSense or Ubuntu, support qemu-kvm, and they run Windows 8.1 very quickly. AES support gives them an edge. Would make great NAS boxes. Their cpu cache (L1, L2) speeds are faster than the AM2 line of cpus, but that are less expensive and less energy used. Great job AMD!
Score
0
Isaiah4110
April 9, 2014 4:42:10 PM
compdoc
April 9, 2014 4:59:01 PM
I think the speed race was won by Intel, but the high end market is limited compared to the huge markets opening in China and Africa, and places where people are being able to afford a computer for the first time in their lives. 3rd World countries will be comparing these low cost apu and atom systems to ARM and Qualcomm and Raspberry Pi. And sure, they are perfect for HTPC. Along routers and mini-servers, that should be a decent size market as well.
Score
0
JimmiGilbi
April 9, 2014 5:17:37 PM
Interesting, I wonder what board I can buy for-
http://www.newegg.ca/Product/Product.aspx?Item=N82E1681...
http://www.newegg.ca/Product/Product.aspx?Item=N82E1681...
Score
0
jlwtech
April 9, 2014 5:51:13 PM
ferooxidan
April 9, 2014 6:03:30 PM
!
, dont get me wrong, my Athlon750k is decent, but could be better